WebLogic load balancing

Question

I'm currently developing a project supported on a WebLogic clustered environment. I've successfully set up the cluster, but now I want a load-balancing solution (currently, only for testing purposes, I'm using WebLogic's HttpClusterServlet with round-robin load-balancing). Is there any documentation that gives a clear comparison (with pros and cons) of the various ways of providing load-balancing for WebLogic?

These are the main topics I want to cover:

• Performance (normal and on failover);
• What failures can be detected and how fast is the failover recovery;
• Transparency to failure (e.g., ability to automatically retry an idempotent request);
• How well is each load-balancing solution adapted to various topologies (N-tier, clustering)

Thanks in advance for your help.

Solution

Is there any documentation that gives a clear comparison (with pros and cons) of the various ways of providing load-balancing for WebLogic?

It's not clear what kind of application you are building and what kind of technologies are involved. But...

You will find useful information in Failover and Replication in a Cluster and Load Balancing in a Cluster (also look at Cluster Implementation Procedures) but, no real comparison between the different options, at least not to my knowledge. But, the choice isn't that complex: 1. Hardware load balancers will perform better than software load balancers and 2. If you go for software load balancers, then WebLogic plugin for Apache is the recommended (by BEA) choice for production. Actually, for web apps, its pretty usual to put the static files on a web server and thus to use the Apache mod_wl plugin. See the Installing and Configuring the Apache HTTP Server Plug-In chapter.

These are the main topics I want to cover:

• Performance (normal and on failover): If this question is about persistent session, WebLogic uses in memory replication by default and this works pretty well with a relatively low overhead.

• What failures can be detected and how fast is the failover recovery: It is unclear which protocols you're using. But see Connection Errors and Clustering Failover.

• Transparency to failure (e.g., ability to automatically retry an idempotent request): Clarifying the protocols you are using would make answering easier. If this question is about HTTP requests, then see Figure 3-1 Connection Failover.

• How well is each load-balancing solution adapted to various topologies (N-tier, clustering): The question is unclear and too vague (for me). But maybe have a look at Cluster Architectures.

Oh, by the way, another nice chapter that you must read Clustering Best Practices.